Incident Clinical and Mortality Associations of Myocardial Native T1 in the UK Biobank

Background Cardiac magnetic resonance native T1-mapping provides noninvasive, quantitative, and contrast-free myocardial characterization. However, its predictive value in population cohorts has not been studied. Objectives The associations of native T1 with incident events were evaluated in 42,308 UK Biobank participants over 3.17 ± 1.53 years of prospective follow-up. Methods Native T1-mapping was performed in 1 midventricular short-axis slice using the Shortened Modified Look-Locker Inversion recovery technique (WIP780B) in 1.5-T scanners (Siemens Healthcare). Global myocardial T1 was calculated using an automated tool. Associations of T1 with: 1) prevalent risk factors (eg, diabetes, hypertension, and high cholesterol); 2) prevalent and incident diseases (eg, any cardiovascular disease [CVD], any brain disease, valvular heart disease, heart failure, nonischemic cardiomyopathies, cardiac arrhythmias, atrial fibrillation [AF], myocardial infarction, ischemic heart disease [IHD], and stroke); and 3) mortality (eg, all-cause, CVD, and IHD) were examined. Results are reported as odds ratios (ORs) or HRs per SD increment of T1 value with 95% CIs and corrected P values, from logistic and Cox proportional hazards regression models. Results Higher myocardial T1 was associated with greater odds of a range of prevalent conditions (eg, any CVD, brain disease, heart failure, nonischemic cardiomyopathies, AF, stroke, and diabetes). The strongest relationships were with heart failure (OR: 1.41 [95% CI: 1.26-1.57]; P = 1.60 × 10-9) and nonischemic cardiomyopathies (OR: 1.40 [95% CI: 1.16-1.66]; P = 2.42 × 10-4). Native T1 was positively associated with incident AF (HR: 1.25 [95% CI: 1.10-1.43]; P = 9.19 × 10-4), incident heart failure (HR: 1.47 [95% CI: 1.31-1.65]; P = 4.79 × 10-11), all-cause mortality (HR: 1.24 [95% CI: 1.12-1.36]; P = 1.51 × 10-5), CVD mortality (HR: 1.40 [95% CI: 1.14-1.73]; P = 0.0014), and IHD mortality (HR: 1.36 [95% CI: 1.03-1.80]; P = 0.0310). Conclusions This large population study demonstrates the utility of myocardial native T1-mapping for disease discrimination and outcome prediction.

C ardic magnetic resonance (CMR) is the reference standard for evaluation of cardiac structure and function. CMR myocardial native T1 mapping provides quantitative, noninvasive, and contrast-free characterization of myocardial tissue on a pixel-by-pixel basis, comparable to a virtual biopsy of the living heart. 1,2 The clinical utility of myocardial native T1 mapping has been shown in select clinical cohorts, particularly for the diagnosis of acute myocardial injury, myocardial inflammation, myocardial iron overload, Fabry disease, and cardiac amyloidosis. 3 However, the prognostic value of native T1 in large population cohorts without pre-existing disease has not been previously studied.
The UK Biobank is a very large population-based cohort study including detailed CMR and prospective tracking of incident health events through linkages to routine health data. 4 We studied demographic and clinical associations of myocardial native T1 in 42,308 UK Biobank participants. Importantly, we evaluated relationships with key incident diseases and mortality outcomes.  Table 1), as per previous publications using this cohort. 9 We included the following prevalent outcomes: any cardiovascular disease (CVD), any brain disease, valvular heart disease, heart failure, nonischemic cardiomyopathies, cardiac arrhythmias, atrial fibrillation (AF), myocardial infarction (MI), ischemic heart disease (IHD), stroke, hypertension, diabetes, and high cholesterol. Mortality outcomes were defined according to the primary cause of death ascertained from death register data. We considered the following incident events: AF, heart failure, stroke, MI, IHD, all-cause mortality, CVD mortality, and IHD mortality. values in a subset of healthy individuals (n ¼ 19,297), stratified by age and sex. Healthy status was defined as the absence of any CVD or classic vascular risk factors (eg, diabetes, hypertension, high cholesterol, and smoking) at the time of imaging. We took age as recorded at the imaging visit and sex from self-report.
Within the healthy subset, we estimated the association of T1 with age using linear regression models separately for men and women. We observed a sex differential trend of T1 with aging within the healthy subset. Thus, subsequent models are adjusted for age, sex, and age Â sex.
We estimated associations of myocardial native T1 in the entire cohort with prevalent disease and incident events (eg, incident CVDs and mortality outcomes) using logistic regression and Cox proportional hazards regression, respectively. We investigated sex and age differential relationships of the associations with incident diseases and mortality using interaction terms added to models (T1 Â age; T1 Â sex) and stratified analyses by sex and median age where indicated by a significant interaction term. We examined for potential nonlinearity by examining associations in strata of above/below the median T1 value. Prevalent diseases were considered as those present at time of imaging. Incident events were considered as first occurrence of the disease after imaging; that is, individuals with record of an outcome of interest before the index date were excluded from the analysis of that outcome. We selected hematocrit, body mass index (BMI), and heart rate as potential confounders of myocardial native T1, as per previous work. 7 BMI and average heart rate were taken from the imaging visit; hematocrit percentage was measured at baseline recruitment. In secondary analyses, we included additional adjustment for BMI, hematocrit, and heart rate. Effect estimates are expressed as odds ratios (ORs) and HRs per 1 SD increment of T1, and standardized beta coefficients and 95% CIs. We present P values corrected for multiple testing using the Benjamini-Hochberg procedure setting the false discovery rate to 5%. 10 The study is reported in accordance with the STROBE

RESULTS
PARTICIPANT CHARACTERISTICS. Myocardial native T1 was available for 42,894 participants. From these, 586 studies (0.01%) with Dice score <0.7 were excluded.
Over a follow-up period of 3.17 AE 1.53 years, we observed 402 (1.0%) deaths; of these, 76 were attributed to CVD and 44 to IHD ( Table 1). The most common incident diseases were IHD (n ¼ 649, 1.5%) and heart failure (n ¼ 243, 0.6%). There were 241 (1.0%) incident MIs and 215 (0.5%) incident cases each of AF and stroke ( Table 1). Within the healthy subset, women had, on average, higher myocardial native T1 than men across all age groups (range: 44 to 84 years), with the greatest difference at younger ages ( Figure 1, Table 2 Table 2).

ASSOCIATIONS OF MYOCARDIAL NATIVE T1 WITH
PREVALENT DISEASE. Within the entire cohort, higher myocardial native T1 was associated with significantly greater odds of any CVD, any brain disease, heart failure, nonischemic cardiomyopathies, cardiac arrhythmias, AF, stroke, and diabetes (Central Table 3). Of these, the largest effect sizes were observed with heart failure (OR: 1.41 [95% CI:  Table 4). There were  no statistically significant associations of native T1 with incident MI, incident IHD, or incident stroke (Central Illustration, Table 4).

SEX/AGE DIFFERENTIAL PATTERNS AND POTENTIAL
NONLINEARITY. Given that associations with incident disease and mortality outcomes are of the greatest clinical interest, we additionally evaluated age and sex differential dependencies of these relationships (Supplemental Table 3). There was no evidence of a sex differential relationship for any of the incident outcomes. For the mortality outcomes, we observed evidence of a significant interaction of T1 with age.
Thus, for these outcomes, we proceeded to examine associations with T1 stratified by median age (65 years); in doing so, we observed greater magnitude of association in older individuals (Supplemental   Table 5).

THE IMPORTANCE OF POTENTIAL CONFOUNDERS.
Higher BMI and hematocrit were significantly associated with lower myocardial native T1, whereas faster average heart rate was associated with significantly higher native T1 (Supplemental Table 6). In models with additional confounder adjustment, all previously observed associations between higher native T1 and prevalent diseases remained robust, with additional significant relationships observed with valvular heart disease, MI, and IHD (Central  Abbreviations as in Table 1. OD ¼ odds ratio; other abbreviations as in Table 1.  where the reverse is true). We examined these relationships in a population-based cohort, showing association of higher native T1 with key prevalent diseases.
We observed strong associations of higher native T1 with prevalent heart failure and nonischemic cardiomyopathies. In keeping with our findings, Dass et al, 16 Puntmann et al, 17